Structured Query Layer

This example matches a typed path, filters node and edge properties, and returns selected aliases. It also shows query_nl with a fake compiler so the natural language layer stays provider-neutral.

Python

import json
from tonggraph import Graph, query_nl

graph = Graph()
alice = graph.add_node(
    "alice",
    labels=["Person"],
    properties={"name": "Alice", "rank": 3, "active": True, "group": "ai"},
)
bob = graph.add_node(
    "bob",
    labels=["Person"],
    properties={"name": "Bob", "rank": 2, "active": True, "group": "research"},
)
carol = graph.add_node(
    "carol",
    labels=["Person"],
    properties={"name": "Carol", "rank": 4, "active": False, "group": "research"},
)

graph.add_edge(alice, bob, "KNOWS", properties={"note": "team alpha", "weight": 0.8})
graph.add_edge(bob, carol, "KNOWS", properties={"note": "team beta", "weight": 0.6})
graph.add_edge(carol, alice, "KNOWS", properties={"note": "loop"})

spec = {
    "match": [
        {
            "node": "a",
            "labels": ["Person"],
            "where": [
                {"property": "rank", "op": "gte", "value": 2},
                {"property": "group", "op": "in", "value": ["ai", "research"]},
            ],
        },
        {
            "edge": "rel",
            "type": "KNOWS",
            "direction": "out",
            "where": [{"property": "note", "op": "contains", "value": "team"}],
        },
        {"node": "b", "labels": ["Person"], "properties": {"active": True}},
    ],
    "return": ["a", "rel", "b"],
    "limit": 10,
}

def compiler(question, schema):
    assert schema["name"] == "tonggraph_query_dsl_v0"
    assert question == "Which active people are known by team members?"
    return spec

result = {
    "structured": graph.query(spec),
    "snapshot": graph.snapshot().query(spec),
    "natural_language": query_nl(
        graph,
        "Which active people are known by team members?",
        compiler,
    ),
}

print(json.dumps(result, indent=2, sort_keys=True))

Output:

{
  "natural_language": [
    {
      "a": 0,
      "b": 1,
      "rel": 0
    }
  ],
  "snapshot": [
    {
      "a": 0,
      "b": 1,
      "rel": 0
    }
  ],
  "structured": [
    {
      "a": 0,
      "b": 1,
      "rel": 0
    }
  ]
}